# IndicSafe: A Benchmark for Evaluating Multilingual LLM Safety in South Asia

arXiv: https://arxiv.org/abs/2603.17915 · HTML: https://mlsanity.com/paper/2603.17915/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Safety alignment, refusal and guardrails, one of 12 directions in LLM safety and security.
- Builds on 17 of its 32 references that are in the map.
- First posted on arXiv 18 Mar 2026.

## Placed in

- LLM safety and security / [Safety alignment, refusal and guardrails](https://mlsanity.com/safety/safety-alignment-refusal-and-guardrails/)

## Builds on (most cited first, 17 of 17)

- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [A Call for Clarity in Reporting BLEU Scores](https://mlsanity.com/paper/1804.08771/) (2018)
- [RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Models](https://mlsanity.com/paper/2009.11462/) (2020)
- [Red Teaming Language Models with Language Models](https://mlsanity.com/paper/2202.03286/) (2022)
- [Underspecification Presents Challenges for Credibility in Modern Machine Learning](https://mlsanity.com/paper/2011.03395/) (2020)
- [Samanantar: The Largest Publicly Available Parallel Corpora Collection for 11 Indic Languages](https://mlsanity.com/paper/2104.05596/) (2021)
- [Towards Afrocentric NLP for African Languages: Where We Are and Where We Can Go](https://mlsanity.com/paper/2203.08351/) (2022)
- [Ethical Reasoning and Moral Value Alignment of LLMs Depend on the Language we Prompt them in](https://mlsanity.com/paper/2404.18460/) (2024)
- [PARIKSHA: A Large-Scale Investigation of Human-LLM Evaluator Agreement on Multilingual and Multi-Cultural Data](https://mlsanity.com/paper/2406.15053/) (2024)
- [Safer or Luckier? LLMs as Safety Evaluators Are Not Robust to Artifacts](https://mlsanity.com/paper/2503.09347/) (2025)
- [The Scales of Justitia: A Comprehensive Survey on Safety Evaluation of LLMs](https://mlsanity.com/paper/2506.11094/) (2025)
- [Hate Speech Detection from Code-mixed Hindi-English Tweets Using Deep Learning Models](https://mlsanity.com/paper/1811.05145/) (2018)
- [SweEval: Do LLMs Really Swear? A Safety Benchmark for Testing Limits for Enterprise Use](https://mlsanity.com/paper/2505.17332/) (2025)
- [DetoxLLM: A Framework for Detoxification with Explanations](https://mlsanity.com/paper/2402.15951/) (2024)
- [FairEval: Evaluating Fairness in LLM-Based Recommendations with Personality Awareness](https://mlsanity.com/paper/2504.07801/) (2025)
- [IndoSafety: Culturally Grounded Safety for LLMs in Indonesian Languages](https://mlsanity.com/paper/2506.02573/) (2025)
- [IndicJR: A Judge-Free Benchmark of Jailbreak Robustness in South Asian Languages](https://mlsanity.com/paper/2602.16832/) (2026)

## Built on by (most cited first, 3 of 3)

- [Register Shifts Break LLM Safety: A Bengali Benchmark with Culturally Grounded Harms](https://mlsanity.com/paper/2608.22335/) (2026)
- [Safety That Does Not Transfer: Cross-Lingual Clinical Correctness Drift in Deployable Medical Language Models](https://mlsanity.com/paper/2607.17270/) (2026)
- [LLM Safety Alignment in Low-Resource Languages: A Systematic Literature Review](https://mlsanity.com/paper/2608.14626/) (2026)

## Abstract (arXiv)

As large language models (LLMs) are deployed in multilingual settings, their safety behavior in culturally diverse, low-resource languages remains poorly understood. We present the first systematic evaluation of LLM safety across 12 Indic languages, spoken by over 1.2 billion people but underrepresented in LLM training data. Using a dataset of 6,000 culturally grounded prompts spanning caste, religion, gender, health, and politics, we assess 10 leading LLMs on translated variants of the prompt. Our analysis reveals significant safety drift: cross-language agreement is just 12.8%, and textttSAFE rate variance exceeds 17% across languages. Some models over-refuse benign prompts in low-resource scripts, overflag politically sensitive topics, while others fail to flag unsafe generations. We quantify these failures using prompt-level entropy, category bias scores, and multilingual consistency indices. Our findings highlight critical safety generalization gaps in multilingual LLMs and show that safety alignment does not transfer evenly across languages. We release textscIndicSafe, the first benchmark to enable culturally informed safety evaluation for Indic deployments, and advocate for language-aware alignment strategies grounded in regional harms.